traffic detection via opendatacam, open cv, etc

traffic analysis project for opendatacam, open cv etc so you install a jetson nano with ip cam and triangulating micro board and directed antenna to send sec to green and green distance data to any mobile that receives broadcast bluetooth data (write that app for io/android), this is plug&play hw, you need to program. start asap, show progress, should be done within a few months. read doc: [login to view URL] first milestone: i suggest you take a traffic scene, analyze the car s velocities, define the green line distance and send a video showing that. take any vid ca 5m above a traffic light. this algo needs to estimate what distance a car needs to be to make it on green. in principle: let s say the light will go red in 10 sec, now where is the distance line where a car going ie 50km/h max limit makes it on green? 140m, right? so you send out 140m and 10 sec, and every sec you subtract 1 sec and 14m. that works for undisturbed flow, now if there s a disturbance, estimate the distance! that s machine learning. what car at what distance made it last time when disturbance was at similar %age? ie make a table: %, sec = distance, 20 entries, thats it. indivudual car speed isnt important, the greeen distance line shows the distance you can max have to make it on greeen at given congestion% then you can refine: congestion % s velocity can change a bit, but behavior will mostly be the same. this is about probabilities. on crossing with multiple lights, each sender needs to send the direction (left straight right), so mobile will show 2 or 3 data sets: left 5 sec, 90m; straight 20 sec, 120m; right 25 sec, 50m (maybe more congestion on right). first milestone is just for one lane to make it simple. just send a vid where the distance is estimated and seconds and distance line are bluetooth-broadcast to mobile device (android app showing 2 datasets). afterwards i would send the hw to you and we define more milestones for more features.